Introduction
Methods
Study design
Exposures
Outcome
Genetic variants
Analysis
Comparison to other falsification strategies: MR-Egger and MR-PRESSO
Instrumental inequalities | MR-Egger intercept test | MR-PRESSO global test | |
---|---|---|---|
Key assumptions in addition to the IV conditions | Coarsening of continuous exposures does not induce violations of IV conditions [31] | Homogeneity and linearity in associations between instrument, exposure, and outcome [34] | Homogeneity and linearity in associations between instrument, exposure, and outcome [35] |
Example interpretations of the falsification test | (1) A global test of the IV conditions and the coarsening assumption (2) Assuming coarsening of continuous exposures did not induce violations, a test of the IV conditions | (1) A global test of the IV conditions, homogeneity, and linearity assumptions (2) Assuming homogeneity and linearity, a test of the IV conditions | (1) A global test of the IV conditions, homogeneity, and linearity assumptions (2) Assuming homogeneity and linearity, a test of the IV conditions |
Considerations for statistical inference | No consensus currently exists on the optimal approach for statistical inference [40] | Often limited in power to be able detect violations of the instrumental conditions [38] | Possibly limited in power to detect violations of the instrumental conditions if the percentage of horizontal pleiotropic proposed instruments is less than 10%; less powerful to detect violations of the instrumental conditions when pleiotropy is imbalanced [35] |
Applicability to one sample and/or two sample data settings | Presented approach only applicable in the one-sample setting | Has been applied in both one- and two-sample settings, but may produce biased results when applied to one sample setting [37] | Currently only applied to two-sample data settings |
Results
Exposure | Number of proposed instruments | Number of violations of IV inequalities marginally | Maximum value of the inequalities for all SNPs marginally | Violation of IV inequalities for allele score | Maximum value of instrumental inequalities for allele score |
---|---|---|---|---|---|
Vitamin D | 15 | 0/15 | 0.36 | No | 0.55 |
Alcohol consumption | 28 | 0/28 | 0.34 | Yes | 1.11 |
CRP | 528 | 0/528 | 0.26 | Yes | 1.25 |
Triglycerides | 22 | 0/22 | 0.14 | Yes | 1.02 |
HDL-cholesterol | 82 | 0/82 | 0.15 | Yes | 1.05 |
LDL-cholesterol | 402 | 0/402 | 0.51 | Yes | 1.50 |